Hi @Asha1,
There are a few different graph types here to consider which are all very different from each other.
First, ordination plots that you show in an emperor plot. PCA or distanced based versions PCoA are dimension reduction methods that are used for exploration mainly (though you can certainly do some hypothesis testing on them later, ex PERMANOVA, adonis etc.). You can read more on these here for example, in short the distance between dots (samples) represents how similar/dissimilar they are, depending on the matrix beind used.
The second plots, alpha diversity boxplots are probably the most straightforward, the 2 boxplots representing IQR of your 2 groups and the corresponding non-parametric krusakl-wallis test that compares the ranks of those groups.
The beta diversity boxplots are a bit trickier to grasp initially and you are certainly not the first person to be thrown off by them. Those n- values do not represent individual samples but rather the total number of distances within samples of a group. Perhaps a bit more thorough explanation of those in this thread can help.
Unfortunately, since one of your groups only has an n=2, this makes statistically sound comparisons difficult and under-powered. I would suggest accepting the results of these tests with a grain of salt and proceed with your interpretation cautiously.